eMMC erase and write support on RK3528 is somewhat unreliable, sometime
e.g. mmc erase and write commands will fail with an error.
Use the delay line lock value for half card clock cycle, DLL_LOCK_VALUE,
to set a manual xx_TAP_VALUE to fix the unreliable eMMC support.
This is only enabled for RK3528, remaining SoCs still use the automatic
tap value, (DLL_LOCK_VALUE * 2) % 256, same value we configure manually
for RK3528.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Guard execute_tuning with MMC_SUPPORTS_TUNING, otherwise there will be
build failure:
drivers/mmc/exynos_dw_mmc.c:484:10: error: 'struct dm_mmc_ops' has no member
named 'execute_tuning'
.execute_tuning = exynos_dwmmc_execute_tuning,
^~~~~~~~~~~~~~
Cc: Kaustabh Chakraborty <kauschluss@disroot.org>
Acked-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Some boards have SD card connectors where the power rail cannot be switched
off by the driver. However there are various circumstances when a card
might be re-initialized, such as after system resume, warm re-boot, or
error handling. However, a UHS card will continue to use 1.8V signaling
unless it is power cycled.
If the card has not been power cycled, it may still be using 1.8V
signaling. According to the SD spec., the Bus Speed Mode (function group 1)
bits 2 to 4 are zero if the card is initialized at 3.3V signal level. Thus
they can be used to determine if the card has already switched to 1.8V
signaling. Detect that situation and try to initialize a UHS-I (1.8V)
transfer mode.
Signed-off-by: Tanmay Kathpalia <tanmay.kathpalia@altera.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Exynos7870 is documented in upstream dt-schema. Add it in the U-Boot
driver.
Note that here it seems that Exynos7 DW MMC is perfectly compatible with
Exynos7870 DW MMC. It's not always true, especially in SDIO cards where
data from a 64-bit FIFO is read in two 32-bit halves [1]. Since SDIO
isn't used or implemented here, it's oblivious. But upstream's schema
considers that quirk, so that compatible is followed.
Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=7cbe799ac10f [1]
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
SD UHS mode is already supported by the Exynos DW-MMC driver in mainline
Linux. Using that as reference, add support in the U-Boot driver.
The maximum frequency was capped to 200000000, increase it to 208000000,
which is the required frequency for UHS_SDR104, which has the highest
frequency of all UHS modes. Moreover, add UHS_CAPS to host capailities.
These changes allow both host and card to recognize support for all UHS
modes.
SDR104, SDR50, and DDR50 have their own CLKSEL timing values, which
requires the CIU div value to be set in bits 18:16. Move the function
exynos_dwmci_clksel() below exynos_dwmmc_get_ciu_div() so that the
latter is accessible from the former, and add cases for said timing
modes.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
MMC HS200 and HS400 modes are already supported by the Exynos DW-MMC
driver in mainline Linux. Using that as reference, add support in the
U-Boot driver.
The maximum frequency was capped to 50000000, increase it to 200000000,
which is the required frequency for HS200/HS400. Moreover, add
MMC_MODE_HS200 and MMC_MODE_HS400 to host capailities. These changes
allow both host and card to recognize support for HS200/HS400.
This change also includes a new ops function, namely execute_tuning.
Implementing it would mean that we can no longer rely on the default ops
provided by dw_mmc.c, thus a new ops instance is created with proper
fields. The execute_tuning function is modeled after the one available
in the Linux driver.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Disrupting the regulator voltage during ios configuration messes with
the MMC initialization sequence. Move the VQMMC regulator enable/disable
functions to the MMC power cycle function, similar to how its done for
the VMMC regulator.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
In certain cases, the VQMMC regulator may not support certain voltages.
For instance, a VQMMC regulator which supports only up to 2.7V will not
accept 3.3V as an argument. This is unaccounted for, and thus the driver
incorrectly assumes that the voltage is set successfully.
Fetch the return value in a variable and return if it's non-zero.
(-ENOSYS is exempted as it implies that the voltage adjustment
functionality as a whole isn't supported).
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
During a voltage switch command (CMD11, opcode: SD_CMD_SWITCH_UHS18V),
certain hosts tend to stop responding to subsequent commands. This is
addressed by introducing an additional command flag,
DWMCI_CMD_VOLT_SWITCH.
The associated interrupt bit is defined as DWMCI_INTMSK_VOLTSW. This is
set high when a voltage switch is issued, this needs to be waited for
and set to low. Implement the same in the timeout loop. Do note that
since DWMCI_INTMSK_VOLTSW shares the same bit as DWMCI_INTMSK_HTO (bit
10), the interrupt bit needs to be polled for only if the volt switch
command is issued.
DWMCI_CMD_VOLT_SWITCH also needs to be set for subsequent clken commands
after the volt switch. To ensure this, add a boolean member in the host
private struct (herein named volt_switching), which informs if the last
command issued was for volt switching or not.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
The current implementation polls for the DWMCI_CMD register, for the
DWMCI_CMD_START bit to turn off, which indicates that the command has
been completed. The problem with this approach is that it doesn't
address the DWMCI_INTMSK_CDONE bit in the interrupt register,
DWMCI_RINTSTS. As a result, subsequent commands result in timeout errors.
Re-implement the waiting logic by polling for said interrupt status bit
and setting it low if raised.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
In dwmci_setup_bus(), if the requested frequency is equal to the current
frequency, the function is returned, assuming no changes are required in
associated registers.
On certain SD cards, skipping in such situations may result in a timeout
errors during MMC initialization. Due to the lack of documentation, the
cause is unknown, but removing said check seems to fix the issue.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
These commands are required by struct dm_mmc_ops. Any platform specific
driver may use some or all of the functions in their own ops. Make them
accessible by moving the prototype to the dwmmc.h header.
Signed-off-by: Kaustabh Chakraborty <kauschluss@disroot.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
This driver got successfully tested with an upstream device tree
and an Exynos5250. The board in question is samsung-manta
(Google Nexus 10) which we are getting ready for upstream.
For the u-boot port was just this additional compatible string needed.
Signed-off-by: Lukas Timmermann <uboot@timmermann.space>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
In owl_mmc_prepare_data there is a NULL check for the pointer data but
it happens after data has already been dereferenced. Refactor the code
so that the NULL check happens before any code dereferences data.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
In octeontx_mmc_io_drive_setup drive and slew are tested for being less
than 0 but they are declared as uint fields so this test must always
fail. Just remove the test.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
The check for no response expected in octeontx_mmc_send_cmd needs parens
adding for proper interpretation.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
The function cv1800b_set_tap_delay is only called by
cv1800b_execute_tuning. The latter is where we currently have the #if
guard around MMC_SUPPORTS_TUNING, and so with the option disabled we
have an unused function warning. Move the guard to cover both functions.
Signed-off-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
When xPL_MMC_WRITE is set but also xPL_MMC_TINY is set, the function
sd_read_ssr may be unused. Guard it appropriately.
Signed-off-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Convert device tree property parsing to use dev_read_u32(), which
handles endianness and simplifies the code.
Signed-off-by: Tanmay Kathpalia <tanmay.kathpalia@altera.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Currently xilinx_pm_request API supports four u32 payloads. However the
legacy SMC format supports five u32 request payloads and extended SMC
format supports six u32 request payloads. Add support for the same in
xilinx_pm_request API. Also add two dummy arguments to all the callers
of xilinx_pm_request.
The TF-A always fills seven u32 return payload so add support
for the same in xilinx_pm_request API.
Signed-off-by: Naman Trivedi <naman.trivedimanojbhai@amd.com>
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Acked-by: Senthil Nathan Thangaraj <senthilnathan.thangaraj@amd.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/5ae6b560741f3ca8b89059c4ebb87acf75b4718e.1756388537.git.michal.simek@amd.com
Linux kernel .get_temp() callback reports values in millicelsius,
U-Boot currently reports them in celsius. Align the two and report
in millicelsius. Update drivers accordingly. Update callsites that
use thermal_get_temp() as well.
The 'temperature' command now reports temperature in millicelsius
as well, with additional accuracy. This changes command line ABI
slightly.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Quentin Schulz <quentin.schulz@cherry.de>
Reviewed-by: David Zang <davidzangcs@gmail.com>
[trini: Update test/cmd/temperature.c]
Signed-off-by: Tom Rini <trini@konsulko.com>
SoCFPGA updates for v2025.10:
CI: https://source.denx.de/u-boot/custodians/u-boot-socfpga/-/pipelines/27762
This pull request brings a set of updates across SoCFPGA platforms
covering Agilex5, Agilex7, N5X, and Stratix10. The changes include:
* Agilex5 enhancements:
- USB3.1 enablement and DWC3 host driver support
- System Manager register configuration for USB3
- Watchdog timeout increase and SDMMC clock API integration
- dcache handling improvements in SMC mailbox path
- Enable SPL_SYS_DCACHE_OFF in defconfig
* Clock driver improvements:
- Introduce dt-bindings header for Agilex clocks
- Add enable/disable API and EMAC clock selection fixes
- Replace manual shifts with FIELD_GET usage
* DDR updates:
- IOSSM mailbox compatibility check
- Correct DDR calibration status handling
* Device tree changes:
- Agilex5: disable cache allocation for reads
- Stratix10: add NAND IP node
- Enable driver model watchdog
- Enable USB3.1 node for Agilex5
* Config cleanups:
- Simplify Agilex7 VAB defconfig
- Remove obsolete SYS_BOOTM_LEN from N5X VAB config
- Enable CRC32 support for SoCFPGA
- Increase USB hub debounce timeout
Overall this set improves reliability of DDR and cache flows,
adds missing USB and MMC features for Agilex5, and refines clock
and configuration handling across platforms.
This patch set has been tested on Agilex 5 devkit, and Agilex devkit.
Update the driver to enable or disable the SDMMC clock via
clock driver model API instead of doing it in the driver itself.
This allows for scalability of the driver for various SoCFPGA
devices.
Signed-off-by: Alif Zakuan Yuslaimi <alif.zakuan.yuslaimi@altera.com>
Reviewed-by: Tien Fong Chee <tien.fong.chee@altera.com>
Fix a couple of style issues reported by checkpatch.pl:
- Replace `#ifdef CONFIG_MMC_TRACE` with `#if IS_ENABLED(CONFIG_MMC_TRACE)`
to follow the preferred kernel style for config-dependent branches.
- Drop explicit zero initialization of a static variable.
No functional change intended.
Signed-off-by: Bhimeswararao Matsa <bhimeswararao.matsa@gmail.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Greg Malysa <malysagreg@gmail.com> says:
Between 2025.07 and 2025.10 many header dependency chains were improved,
but this exposed implicit header usage in several of our drivers. This
wasn't discovered before or included in the original fixes because our
drivers are not yet used by any mainline-supported boards, so build
tests did not find them. This series addresses the two build failures
I've encountered while rebasing our work onto 2025.10 and continuing to
prepare the next submission of our board files.
Link: https://lore.kernel.org/r/20250903234205.26787-1-malysagreg@gmail.com
As part of the header dependency cleanup between 2025.07 and 2025.10, an
implicit route to obtain SZ_128M from linux/sizes.h was removed. This
adds an explicit reference to linux/sizes.h to fix build failures for
this driver.
Signed-off-by: Greg Malysa <malysagreg@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Remove unnecessary whitespace before '\n' in trace printf
format strings (checkpatch warning QUOTED_WHITESPACE_BEFORE_NEWLINE).
No functional change intended.
Signed-off-by: Bhimeswararao Matsa <bhimeswararao.matsa@gmail.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
In sdhci_iproc_execute_tuning the variable tuning_loop_counter is
unsigned and therefore will always fail the test for it being less than
0. Fix this by changing the variable type to be s8.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
In rockchip_sdhci_execute_tuning the variable tuning_loop_counter is
tested for being less than 0. Ensure that it is a signed type by
declaring it as s8 instead of char.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Avoid unlimited while loop by adding a timeout. The timeout is
calculated based on a minimal throughput of 256 KB/s.
The timeout is set at least to 2 seconds.
Signed-off-by: Christophe Kerello <christophe.kerello@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
In mci_send_cmd the pointer 'data' is optional so guard its use with a
NULL check to prevent any attempt to dereference it when not provided.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Remove duplicate checks on status from mci_data_read and mci_data_write
which are guaranteed to be true as exiting the above do..while loop
above requires that to be so.
This issue was found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
The Allwinner A523 SoC has a slightly changed mod clock, where the P
factor, formerly a shift value, is now a second divider value.
Also the input clock is not PLL_PERIPH0 (600MHz) anymore, but
PLL_PERIPH0_400M (for MMC0/1), so adjust the input rate calculation
accordingly. MMC2 has a different set of parents, so the input clock
is 800 MHz there.
Adjust for all of this.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This is a fix for the sporadic mmc write failure:
mmc write failed
0 blocks written: ERROR
After this happens the emmc will remain in an error state
where subsequent read/writes fail with a timeout.
The mmc driver sends CMD25 - WRITE_MULTIPLE_BLOCK which
can sporadically timeout. When this happens, the mmc driver
aborts the transfer and returns the above error messages.
But the emmc still remains in data transfer mode, since
the timeout was decided by uboot, not by the emmc.
Fix this by sending the STOP_TRANSMISSION command (CMD12)
and waiting for the emmc to be in ready state again (CMD13).
Transferring data blocks after a CMD25 can take anywhere
between 5 and +15s on Samsung EMMCs and the current timeout
is not enough. Increase the timeout by 2x to accommodate the
long transfer times observed.
Signed-off-by: Iulian Banaga <iulianeugen.banaga@mobileye.com>
Acked-by: Jimmy Durand Wesolowski <jimmy.wesolowski@mobileye.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
The exynos5420 DTSes in linux kernel uses the compatibles
samsung,exynos5420-dw-mshc{,-smu} instead of just
samsung,exynos-dwmmc. Match the additional compatibles in the driver
to make it possible to use it with DTSes from Linux kernel.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
[ grimler: rebase after clksel reg abstraction and re-write commit message ]
Signed-off-by: Henrik Grimler <henrik@grimler.se>
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Anand Moon <linux.amoon@gmail.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
the Exynos4210 devices in u-boot uses samsung,exynos4412-sdhci as
sdhci compatible in their DTSes, while the upstream DTSes uses
samsung,exynos4210-sdhci.
Add samsung,exynos4210-sdhci compatible string for s5p_sdhci driver as
well so that it can be used with upstream DTSes.
Signed-off-by: Henrik Grimler <henrik@grimler.se>
Reviewed-by: Anand Moon <linux.amoon@gmail.com>
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
The common sdhci infrastructure is already parsing the cd-gpios
property, since commit 451931ea7002 ("mmc: sdhci: Read cd-gpio from
devicetree"). The s5p code is therefore duplicated, and also broken
since it assumes that the GPIO value is inverted, while the sdhci code
correctly follows the ACTIVE_LOW/ACTIVE_HIGH flag specified in the
device tree.
This fix was originally authored by Simon Shields:
https://github.com/fourkbomb/u-boot/commit/2eac9dea7903
The change has been tested on exynos4210-i9100, a device similar to
exynos4210-trats.
Signed-off-by: Henrik Grimler <henrik@grimler.se>
Reviewed-by: Anand Moon <linux.amoon@gmail.com>
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
A large number of mmc drivers cannot build without access to some
platform specific header files. Express those requirements in Kconfig as
well.
Signed-off-by: Tom Rini <trini@konsulko.com>
Acked-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Instead of returning -EINVAL directly which will not call the cleanup
path to free memory, fix the code to set the error and then goto the
cleanup code.
This issue found by Smatch.
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Reset the host controller at the early stage of probe so that the
configuration will be done properly for reboot cases.
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Link: https://lore.kernel.org/r/20250626062440.295301-1-venkatesh.abbarapu@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Update the MMC driver to retrieve the clock manager base address via
probing the clock manager node in the device tree.
Signed-off-by: Alif Zakuan Yuslaimi <alif.zakuan.yuslaimi@altera.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
This reverts commit ae93d8106bdb5926efef9222d553adb295ebce96.
It is no longer needed since 24b1e0c7e2e3.
Since the obsolete include pulled in byteorder.h which is needed by now,
include this one directly.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Marek Vasut <marek.vasut+renesas@mailbox.org> says:
Rename the environment related variables and add ENV_ prefix, so that
all configuration options which are related to environment would have
an CONFIG_ENV_ prefix. No functional change.
Link: https://lore.kernel.org/r/20250609192701.20260-1-marek.vasut+renesas@mailbox.org
